Hive Key Features
Familiar SQL-like Interface:
Use existing SQL skills to run batch queries on data stored in Hadoop. Queries are written using a SQL-like language, HiveQL, and are executed through either MapReduce or Apache Spark™, making it simple for more users to process and analyze unlimited amounts of data.
Shared Data Structures:
Using HCatalog, a table and storage management layer for Hadoop, Hive metadata is exposed to other data processing tools, including Pig and MapReduce, as well as through a REST API. This allows users to easily read and write data without worrying about where the data is stored, what format it is, or redefining the structure for each tool.
Faster Batch Processing:
Hive-on-Spark features the next generation of batch processing for Hive. With queries executed through Apache Spark™, a powerful data processing tool, users will see dramatic performance improvements compared to MapReduce.
Common Use Cases
Integrated across the platform
As an integrated part of Cloudera’s platform, users can run batch processing workloads with Apache Hive, while also analyzing the same data for interactive SQL or machine-learning workloads using tools like Impala or Apache Spark™ — all within a single platform.
Hive also benefits from unified resource management (through YARN), simple deployment and administration (through Cloudera Manager), and shared compliance-ready security and governance (through Apache Sentry and Cloudera Navigator) —- all critical for running in production.
The shift to Hive-on-Spark
Apache Spark™ is a powerful data processing engine that has quickly emerged as an open standard for Hadoop due to its added speed and greater flexibility. Together with the community, Cloudera has been working to evolve the tools currently built on MapReduce, including Hive and Pig, and migrate them to the Spark execution engine for faster processing.